Aimee S. Dunlap

Components of change in the evolution of learning and unlearned preference

Several phenomena in animal learning seem to call for evolutionary explanations, such as patterns of what animals learn and do not learn. While several models consider how evolution should influence learning, we have very little data testing these models. Theorists agree that environmental change is a central factor in the evolution of learning. We describe a mathematical model and an experiment, testing two components of change: reliability of experience and predictability of the best action. Using replicate populations of Drosophila we varied statistical patterns of change across 30 generations. Our results provide the first experimental demonstration that some types of environmental change favour learning while others select against it, giving the first experimental support for a more nuanced interpretation of the selective factors influencing the evolution of learning.

Experimental evolution of prepared learning

Significance Learning is one of the most basic phenomena in the behavioral sciences. Animals learn some things better than others, and understanding what constrains this basic process is fundamental to our understanding of learning. Our paper applies an evolutionary approach to this question. We offer a simple model that considers the fitness of value of “prepared learning,” and we test this model using experimental evolution. In doing so, we created different lines of Drosophila that are prepared to learn from different experiences. To the best of our knowledge this is the first mathematical model explaining why some associations are learned more easily than others and to our knowledge is the first time that the evolution of prepared learning has been demonstrated experimentally. Animals learn some things more easily than others. To explain this so-called prepared learning, investigators commonly appeal to the evolutionary history of stimulus–consequence relationships experienced by a population or species. We offer a simple model that formalizes this long-standing hypothesis. The key variable in our model is the statistical reliability of the association between stimulus, action, and consequence. We use experimental evolution to test this hypothesis in populations of Drosophila. We systematically manipulated the reliability of two types of experience (the pairing of the aversive chemical quinine with color or with odor). Following 40 generations of evolution, data from learning assays support our basic prediction: Changes in learning abilities track the reliability of associations during a population’s selective history. In populations where, for example, quinine–color pairings were unreliable but quinine–odor pairings were reliable, we find increased sensitivity to learning the quinine–odor experience and reduced sensitivity to learning quinine–color. To the best of our knowledge this is the first experimental demonstration of the evolution of prepared learning.

Experimental Evolution and Economics

This is a theory paper that advocates experimental evolution as a novel approach to study economic preferences. Economics could benefit because preferences are exogenous, axiomatic, and contentious. Experimental evolution allows the empirical study of preferences by placing organisms in designed environments and studying their genotype and phenotype over multiple generations. We describe a number of empirical studies on different aspects of preferences. We argue that experimental evolution has the potential to improve economics.

Why do animals make better choices in patch-leaving problems?

This study compares two procedures for the study of choices that differ in time and amount, namely the self-control and patch procedures. The self-control procedure offers animals a binary mutually exclusive choice between a smaller-sooner and larger-later option. This procedure dominates the choice literature. It seems to address the idea of choice in a general, but relatively abstract way. Animals in the self-control situation frequently prefer the smaller-sooner option even when the larger-later option yields a higher long-term intake rate. In contrast, the patch procedure poses an economically similar question, but simulates the naturally occurring problem of patch exploitation. In the patch procedure, animals choose between leaving and staying. Emerging evidence suggests that animals perform better and achieve higher long-term intake rates in the patch situation. This observation raises the question of how a single set of choice mechanisms could produce these different outcomes. The experiment presented here tests two hypotheses about the relationship between the patch and self-control situations. First, it asks whether the short-term rate rule can predict choice behavior in both situations. Second, it tests the second-delivery hypothesis which holds that the patch situation favors choosing the larger more delayed option (staying) because this option ultimately leads to two food deliveries. The results of this experiment convincingly reject both of these hypotheses. Indeed, our results suggest that none of the simple rules based on time and amount can explain the observed differences between the patch and self-control situations. This result challenges the generality of existing models of choice.

Patch exploitation as choice: symmetric choice in an asymmetric situation?

In the present paper, we explore a novel preparation for the study of animal choice behaviour designed to capture some aspects of naturally occurring patch exploitation. Although one can cast the problem of patch exploitation as a binary choice, naturally occurring patch-leaving decisions are inevitably asymmetric. We asked whether captive blue jays, Cyanocitta cristata , treat leaving and staying in the same way. To do this we factorially varied the delays associated with leaving and staying in a food patch. In addition, we manipulated our subject’s level of motivation (e.g. hunger) using prefeeding treatments. We found that hungry subjects came closest to our prediction of treating leaving and staying in the same way, but that less motivated subjects showed a pronounced and surprising bias in favour of leaving. We discuss the implications of results for experimental and theoretical studies and choice behaviour. We suggest that students of choice behaviour need to understand the sources of such biases because naturally occurring choice situations are seldom perfectly symmetrical.

Components of change and the evolution of learning in theory and experiment

Theoretical treatments of the evolution of learning have a long and rich history, and although many aspects remain unresolved, the consensus is that the predictability and timescale of environmental change play a crucial role in when learning evolves. Directly testing these ideas has proven difficult because comparative experiments must assume many often unknowable aspects of an evolutionary past. Even within the present, identifying and accurately quantifying the relevant types of change can be problematic. Controlling or manipulating change can be difficult in many taxa. Within the theory, what is meant by change can markedly vary between models. Here, we present a targeted comparison of models to show this variation, and argue that standardizing measures of change can add tractability to models. We first review how change is emphasized in models of learning evolution and then describe the still small literature that directly tests the evolution of learning via digital evolution and experimental evolution. We then give an example of how to tie specific natural history to larger theory on learning evolution using the flag model of reliability and certainty and foraging in bumblebees. Learning, by its nature, is of fundamental importance to many fields. Theoretical treatments of learning evolution have been growing at a rapid pace, often with limited empirical applicability to natural systems and little congruence on what is meant by change across models. By explicitly defining change and tying models to natural systems, we can greatly increase our ability to not only understand when learning should evolve, but also when learning does evolve.